Every day, courts, authorities, and public institutions create documents packed with insights – yet most of this knowledge remains locked away. Why? Because sensitive data can’t be shared freely.
That’s where automated anonymization comes in. Together with the Austrian Federal Ministry of Justice (BMJ) and the Federal Computing Center (BRZ), EBCONT has developed a solution that safely unlocks this data treasure – improving processes, preventing mistakes, and enabling smarter decisions, all while keeping privacy intact.
The starting point: valuable data behind high walls
Court rulings contain a wealth of valuable information – from reasoning and interpretations of the law to references to similar cases. At the same time, they are full of sensitive data such as names, addresses, or dates of birth. Without anonymization, these documents can hardly be shared or analyzed.
Until now, this meant painstaking manual processing for the judiciary, a variety of formats to manage, and a high risk of errors in complex cases. The Austrian Federal Ministry of Justice (BMJ) has therefore set the goal of making access to this information faster and more secure – while maintaining the highest standards of data protection.
Why standard AI falls short
Legal texts are tricky: technical terms get misinterpreted, regional nuances slip through, and context is often lost. That’s why the BMJ teamed up with EBCONT to develop custom AI models tailored to Austrian legal language.
The result? Documents are automatically anonymized – names, addresses, case numbers, and roles – while staying fully readable. Using a hybrid mix of NER, deep learning, rule-based methods, and OCR, and hosted on-premise, the solution keeps data secure, GDPR-compliant, and independent of US-based clouds. Accuracy? Over 90%.
Continuous improvement as a success factor
With their complexity and numerous special cases, legal texts can be difficult to navigate. To keep the system flexible, it combines rule-based methods with machine learning, is continuously retrained, and optimized through feedback from practical use. This ensures the solution remains up-to-date and reliable, even as requirements change.
With this new solution, the BMJ can process court decisions more quickly and efficiently – while making them accessible to a wider audience without compromising data protection. The benefits are clear: faster processes, GDPR-compliant use, better-structured data for analysis, and new digital services.
The concept is also transferable – from ministries to police forces to the healthcare sector. The platform is designed to grow continuously, for example by adding audio transcriptions, multilingual translations, or interactive queries using generative AI. In this way, the anonymization solution evolves into a comprehensive data platform for the judiciary – and a model from which other authorities can also benefit.
The project demonstrates that sensitive data can not only be protected but also become a driver for digitalization and innovation – provided that data protection, expertise, and technology work hand in hand.
Sensitive data: an opportunity, not a risk
Frederick Bednar, Data Analytics and Data Science Consultant at EBCONT, says: “The BMJ project shows that sensitive data does not have to conflict with digitalization – on the contrary: when data protection, technology, and expertise work together, they become a real competitive advantage.”
For decision-makers in the public sector, this means: those who prioritize data sovereignty and tailored AI solutions can accelerate processes, drive innovation, and build trust – without compromising on security or compliance.